Title
Author(s)
Citation
Issued Date
URL
Rights
Hand, foot, and mouth disease in China, 2008-12: an
epidemiological study
Xing, W; Liao, Q; Viboud, C; Zhang, J; Sun, J; Wu, JTK; Chang,
Z; Liu, F; Fang, J; Zheng, Y; Cowling, BJ; Varma, JK; Farrar, JJ;
Leung, GM; Yu, H
The Lancet Infectious Diseases, 2014, v. 14 n. 4, p. 308-318
2014
http://hdl.handle.net/10722/196741
NOTICE: this is the author’s version of a work that was accepted
for publication in The Lancet Infectious Diseases. Changes
resulting from the publishing process, such as peer review,
editing, corrections, structural formatting, and other quality
control mechanisms may not be reflected in this document.
Changes may have been made to this work since it was
submitted for publication. A definitive version was subsequently
published in The Lancet Infectious Diseases, 2014, v. 14 n. 4, p.
308-318. DOI: 10.1016/S1473-3099(13)70342-6; This work is
licensed under a Creative Commons AttributionNonCommercial-NoDerivatives 4.0 International License.
1
Epidemiological characteristics of hand-foot-and-mouth disease in
2
China, 2008-2012
3
4
Weijia Xing*, Qiaohong Liao*, Cécile Viboud*, Jing Zhang*, Junling Sun, Joseph T
5
Wu, Zhaorui Chang, Fengfeng Liu, Vicky J Fang, Yingdong Zheng, Benjamin J
6
Cowling, Jay K Varma, Jeremy J Farrar, Gabriel M Leung, Hongjie Yu
7
8
Division of Infectious Disease, Key Laboratory of Surveillance and
9
Early–warning on Infectious Disease, Chinese Center for Disease Control and
10
Prevention, Beijing, China (W Xing PhD, Q Liao MD, J Zhang MD, J Sun PhD,
11
Zhaorui Chang MD, F Liu MD, H Yu MD); Fogarty International Center, National
12
Institutes of Health, Bethesda, MD, USA (C Viboud PhD); Division of
13
Epidemiology and Biostatistics, School of Public Health, Li Ka Shing Faculty of
14
Medicine, The University of Hong Kong, Hong Kong Special Administrative
15
Region, China (B J Cowling PhD, J T Wu PhD, V J Fang MPhil, Prof G M Leung
16
MD); School of Public Health, Peking University, Health Science Center, Beijing,
17
China (Y Zheng PhD); New York City Department of Health and Mental Hygiene,
18
New York, USA (J K Varma MD); Oxford University Clinical Research Unit –
19
Wellcome Trust Major Overseas Programmes, Vietnam (Prof J J Farrar PhD);
20
ISARIC, Centre for Tropical Medicine, University of Oxford, Churchill Hospital,
1
1
Oxford, United Kingdom (Prof J J Farrar PhD); Department of Medicine, National
2
University of Singapore, Singapore (Prof J J Farrar PhD); The Li Ka Shing Oxford
3
Global Health Programme and Wellcome Trust (Prof J J Farrar PhD)
4
*
5
Correspondence to:
6
Prof Gabriel M Leung, Li Ka Shing Faculty of Medicine, The University of Hong
7
Kong, 21 Sassoon Road, Hong Kong SAR, China
8
[email protected]
9
or
Contributed equally
10
Dr Hongjie Yu, Division of Infectious Disease, Key Laboratory of Surveillance and
11
Early–warning on Infectious Disease, Chinese Center for Disease Control and
12
Prevention, No. 155 Changbai Road, Changping district, Beijing, China
13
[email protected]
14
2
1
Summary
2
3
Background Hand–foot–and–mouth disease (HFMD) is a common childhood illness
4
caused by enteroviruses. Increasingly it imposes a substantial disease burden
5
throughout East and Southeast Asia. To better inform vaccine and other interventions,
6
we characterized the epidemiology of HFMD in China based on enhanced
7
surveillance.
8
Methods We extracted epidemiological, clinical and laboratory data from reported
9
HFMD cases during 2008–2012 and compiled climatic, geographic and demographic
10
information. All analyses were stratified by age, disease severity, laboratory
11
confirmation status and enterovirus subtype.
12
Findings The surveillance registry captured 7,200,092 probable HFMD cases
13
(annualized incidence, 1·2 per 1,000), of whom 3·7% were laboratory–confirmed and
14
0·03% died. Incidence and mortality were highest in children aged 12–23 months (in
15
2012: 38·2 cases per 1,000 and 1·5 death per 100,000). Median durations from onset
16
to diagnosis and death were 1·5 days and 3·5 days respectively. The risk of
17
cardiopulmonary or neurological complications was 1·1% and the severe-case fatality
18
risk was 3·0%, with >90% of deaths associated with enterovirus 71. HFMD peaked
19
annually in June in the North, whereas Southern China experienced semi-annual
20
outbreaks in May and September/October. Geographic differences in seasonal
21
patterns were weakly associated with climate and demographic factors (variance
3
1
explained 8-23% and 3–19%, respectively).
2
Interpretation This is the largest population-based study to date of the epidemiology
3
of HFMD. Future mitigation policies should take full account of the heterogeneities of
4
disease burden identified. Additional epidemiologic and serologic studies are
5
warranted to elucidate local HFMD dynamics and immunity patterns and optimize
6
interventions.
7
Funding China–US Collaborative Program on Emerging and Re-emerging Infectious
8
Diseases; World Health Organization; The Li Ka Shing Oxford Global Health
9
Programme and Wellcome Trust; Harvard Center for Communicable Disease
10
Dynamics; Health and Medical Research Fund, Government of the Hong Kong
11
Special Administrative Region.
12
13
Key words Hand, Foot and Mouth Disease; Enterovirus 71; Coxsackie virus A16;
14
Epidemiology; Disease burden; Seasonality
15
16
Word count: text 3197; abstract: 260
4
1
Introduction
2
Hand-foot-and-mouth disease (HFMD) is a common infectious condition caused by a
3
variety of enteroviruses, with enterovirus 71 (EV-A71) and Coxsackie virus A16
4
(CA-V16) being the most commonly reported.1 Children under five years are
5
particularly prone to HFMD, and most patients show self-limiting illness typically
6
including fever, skin eruptions on hands and feet, and vesicles in the mouth. However,
7
a small proportion has been known to rapidly develop neurological and systemic
8
complications that can be fatal, particularly in cases associated with EV-A71.2
9
EV-A71 was first identified in 1969 in California, USA3 and has become more
10
predominant across the Asia–Pacific region in the past 15 years.4̶–11
11
China established a national enhanced surveillance system for HFMD in May 2008
12
partly in response to several large HFMD outbreaks during 2007 and early 2008.9
13
Here we describe the enhanced HFMD surveillance system and characterize the
14
epidemiology of the disease in China, focusing on age, seasonal, and geographic
15
patterns from 2008 to 2012.
16
17
Methods
18
Enhanced national surveillance program
19
Beginning on January 1, 2008, all probable and laboratory–confirmed HFMD cases
20
were reported on a voluntary basis to the Chinese Centre for Disease Control and
5
1
Prevention (China CDC) in Beijing. On May 2, 2008, HFMD was made statutorily
2
notifiable.
3
4
Case definitions
5
A probable HFMD case was defined as a patient with papular or vesicular rash on
6
hands, feet, mouth, or buttocks, with or without fever. A confirmed case was defined
7
as a probable case with laboratory evidence of enterovirus infection (including
8
EV-A71, coxsackievirus A16 (CA-V16), or other non-EV-A71 and non-CA-V16
9
enteroviruses) detected by reverse-transcriptase polymerase chain reaction (RT-PCR),
10
real-time RT-PCR, or virus isolation.
11
HFMD patients, whether probable or confirmed, were classified as severe if they
12
experienced any neurological complications (aseptic meningitis, encephalitis,
13
encephalomyelitis, acute flaccid paralysis, or autonomic nervous system dysregulation)
14
and/or cardiopulmonary complications (pulmonary oedema, pulmonary haemorrhage,
15
or cardiorespiratory failure); otherwise, they were categorized as mild cases.
16
17
Collection and testing of specimens
18
Given limited laboratory capacity during the first year of enhanced surveillance (from
19
May 2008 through June 2009), clinical samples were collected from the first five mild,
20
probable cases who visited hospital outpatient departments each week in each of 31
6
1
Chinese provinces (Appendix Figure 1). We also attempted to collect specimens from
2
all severe cases. With enhanced capacity after June 2009, samples were collected from
3
the first five mild, probable cases who visited hospital outpatient departments each
4
month in each of the 3,074 counties or districts and from all severe cases (Appendix
5
Figure 1).12 Depending on the symptoms and clinical status of each case, the
6
appropriate clinical specimens, including throat swab, rectal swab, fecal sample,
7
vesicular fluid, and/or cerebrospinal fluid, were collected. Sampling was systematic
8
and did not follow community outbreaks.
9
Specimens were placed in sterile viral transport medium and sent to provincial– or
10
prefecture–level CDCs (n=425) for PCR or virus isolation, according to standardized
11
protocols disseminated by the national CDC.13 Viral ribonucleic acid (RNA) was
12
extracted using available commercial kits (most frequently, QIAamp Viral RNA Mini
13
Kit, Qiagen, Valencia, CA, USA, and Geneaid Viral RNA Mini Kit, Geneaid Biotech,
14
Taiwan China) as per the manufacturer’s protocols. RNA from each sample was tested
15
for specific primers and probes that targeted pan-enterovirus, EV-A71, and CA-V16.
16
Testing was carried out in biosafety level 2 facilities. Test results were classified into
17
four categories: enterovirus negative, EV-A71 positive, CA-V16 positive, or positive
18
for another enterovirus without further serotype identification.
19
A small number of samples underwent virus isolation at provincial–level CDCs.
20
According to national guidelines,12 at least ten enterovirus strains should be identified
7
1
in each province every month. Specimens were inoculated into human
2
rhabdomyosarcoma (susceptible to EV-A71 and CA-V16) and Hep-2 (susceptible to
3
other enteroviruses) cells. If cultures showed cytopathic effect (CPE), serotype
4
identification was obtained by sequencing analysis14. A blind passage was done with
5
all cultures showing no CPE after 7 days.
6
7
Clinical and epidemiologic data
8
Probable and confirmed case were reported on-line to the national CDC within 24
9
hours of diagnosis, using a standardized form including basic demographic
10
information (sex, date of birth, and address), case classification (probable or
11
confirmed), severity (mild or severe), death status, date of symptoms onset, date of
12
diagnosis, and date of death (if applicable), and virus type (EV-A71, CA-V16, other
13
enterovirus) for confirmed cases.
14
15
Climate, geographic, and demographic data
16
To analyze the potential drivers of HFMD seasonality, we collected demographic,
17
economic, geographic, and climatic information for all provinces from 2008 to 2012,
18
including: 1) age–specific population denominators, population density, Gross
19
Domestic Product, and Gross Regional Product;15 2) latitude and longitude of
20
province capitals; 3) air, rail, road, and boat passenger data;16 4) meteorological data
8
1
including daily temperature, rainfall, relative humidity, air pressure, average vapor
2
pressure, and hours of sunshine (Appendix Text 1 for details).17 Climate variables
3
were aggregated by season (spring: April–June; autumn: September–November), to
4
reflect the seasonal occurrence of HFMD peaks in surveillance data. The 31 provinces
5
were further grouped into six climatic regions (Appendix Figure 2A).
6
7
Data analysis
8
Disease burden and severity of disease
9
We included all cases with illness onset from January 1, 2008 through December 31,
10
2012 in the analysis. We estimated age–specific rates of incidence, severe illness, and
11
mortality by combining probable and confirmed cases. 95% confidence intervals (CIs)
12
for rates were estimated with Poisson methods. For rates with denominator >100,000
13
and numerator ≥20, we calculated 95% CIs assuming a normal distribution. We
14
assessed the geographic distribution of cases across all 31 provinces, grouped into
15
seven regions as shown in Appendix Figure 2B. We estimated the age specific
16
case–severity risk (no. of severe cases/no. of probable and confirmed cases),
17
case–fatality risk (no. of deaths/no. of probable and confirmed cases), and
18
case–fatality risk of severe HFMD cases (no. of deaths/no. of severe cases), overall
19
and stratified by serotype. In serotype-specific analyses, we estimated the number of
20
serotype-specific HFMD cases by applying the distribution of serotypes among
9
1
specimens positive for enteroviruses and further serotyped to the universe of probable
2
and confirmed cases.
3
We calculated the distributions of times from illness onset to diagnosis, illness onset
4
to death, and diagnosis to death by virus serotype. To identify predictors for severe or
5
fatal outcomes, we applied logistic regression and backward selection of demographic
6
and geographic variables, time from onset to diagnosis, and virus type (p-value for
7
exclusion≥0·10), separately for laboratory–confirmed cases and probable cases. All
8
analyses were carried out using SAS version 9.1 (SAS Institute, Cary, USA).
9
Geographic patterns and seasonality
10
To quantify HFMD seasonal patterns by province, we created heatmaps of the
11
proportion of HMFD cases identified in each week of the year.18 We also fitted
12
seasonal multiple linear regression models to weekly HFMD time series, including
13
time trends and harmonic terms representing annual and semi-annual periodic cycles
14
(Appendix Text 2).18,19 We extracted the amplitude and peak timing of the annual and
15
semi-annual cycles based on model coefficients.18 The amplitude measures the
16
difference between the maximum and minimum of a seasonal curve.19
17
We used multivariate stepwise linear regression to study the association between
18
HFMD seasonal characteristics and putative seasonal drivers, including geography,
19
human mobility patterns, demographic, economic, and climate variables.18
20
Epidemiologically relevant HFMD regions were identified by hierarchical clustering
10
1
using Ward’s minimum variance method, and predictors of these regions were
2
identified through discriminant analysis.18 Time series were standardized for
3
differences in sampling intensity over time and geography by dividing weekly case
4
counts by the annual number of cases.
5
6
Results
7
Incidence description
8
7,200,092 probable HFMD cases were reported to the China CDC surveillance system
9
during 2008–2012, of which 3·7% were laboratory–confirmed and 0·03% died. The
10
incidence of reported HFMD cases showed a sharp increase since the initiation of
11
surveillance in 2008 (Table 1), due to improvements in reporting and surveillance.
12
Reported disease rates reached a steady state at over 1·2 cases per 1,000 person–years
13
during 2010-2012.
14
HFMD incidence varied greatly with age, with highest rates in children aged six
15
months to two years (Table 1). The median age of reported cases was 27 months
16
(interquartile range = 18–43 months). Most cases occurred in children below the age
17
of five, and incidence was very low in young infants aged <6 months, older children,
18
and adults. The incidence of HFMD was 1·6 times higher in boys under 5 years than
19
in girls of the same age (p<0·001).
20
EV-A71 predominated among laboratory-confirmed cases, accounting for 45%, 80%,
11
1
and 93% of mild, severe, and fatal cases respectively (Figure 1). There was little
2
variation in EV-A71 predominance by age (Figure 2A). EV-A71, CA-V16, and other
3
enteroviruses co-circulated throughout the observation period in all regions
4
(Appendix Figure 3).
5
6
Clinical course
7
The median time from illness onset to diagnosis, illness onset to death, and diagnosis
8
to death were 1·5 days (interquartile range: 0·5–2·5 days), 3·5 days (interquartile
9
range: 2·5–4·5 days), and 0·5 day (interquartile range: 0·5–1·5 days) respectively.
10
The probability density distributions of the onset-to-diagnosis, onset-to-death, and
11
diagnosis-to-death intervals were similar between probable and laboratory–confirmed
12
cases of EV-A71 or other enteroviruses (Figure 3). However, CA-V-16 density
13
distributions showed attenuated signal-to-noise ratio given the small number of deaths
14
(n=25).
15
16
Severity of disease
17
Overall the case–fatality, case–severity, and severe case–fatality risks were 0·03%
18
(range across years = 0·03–0·05%), 1·1% (range = 0·2–1·6%), and 3·0% (range =
19
2·6–10·4%) respectively. Illness severity and death were inversely related to age
20
(Table 2 and Figure 2B–2D). Consistently across all age groups, EV-A71 infections
12
1
were much more severe than CA-V16 infections (Figure 2B–2D). We performed
2
sensitivity analyses on severity estimates by age and viral etiology limited to
3
2010-2012 (Appendix Figure 4) and only 2012 (Appendix Figure 5), and found
4
results were similar. Multivariate backward logistic regression of laboratory
5
confirmed cases identified several mortality risk factors beyond young age and
6
infection with EV-A71, including rural residence and long onset-to-diagnosis interval.
7
Further, males and rural residents experienced higher risk of severe disease. Including
8
probable cases in the analysis yielded similar risk predictors (Table 2).
9
10
Seasonal characteristics
11
National and province-specific patterns
12
At the national scale, HFMD showed semi-annual peaks of activity, including a major
13
peak in spring and early summer followed by a smaller peak in autumn, consistent
14
between probable and confirmed time series (Figure 4).
15
While Southern China experienced two outbreaks of HFMD peaking in May and
16
October of each year, Northern China experienced a single annual peak in June
17
(Figure 5-6). The annual amplitude of HFMD epidemics increased with increasing
18
latitude and semi-annual periodicity was strongest in the South (Figure 5).
19
Seasonal characteristics by serotype
20
All enteroviruses exhibited latitudinal gradients in the amplitude of annual and
13
1
semi-annual epidemics (Appendix Figures 6–8). Annual peak timing of CA-V16
2
activity occurred earlier than for other enteroviruses (p<0·001).
3
Predictors of HFMD seasonality
4
Putative predictors of HFMD seasonal characteristics were identified through
5
multivariate stepwise regression (Appendix Table 1). We found that the annual
6
amplitude of epidemics was associated with spring rainfall and spring sunshine in
7
overall and serotype–stratified analyses (p<0·05, partial R2: 5%–23%; p<0·05,
8
8%–23%, respectively). HFMD peak timing was associated with maximum spring air
9
pressure in overall and serotype–stratified analyses (p<0·05, partial R2: 4%–12%).
10
Predictors of semi-annual periodicity of HFMD epidemics were more varied.
11
Population and transportation factors were moderately associated with HFMD
12
seasonal characteristics, explaining 1–19% of the variance in epidemic timing and
13
periodicity. Overall, multivariate models explained a moderate-to-high fraction of the
14
variance in HFMD seasonal characteristics through multiple factors (40–92%).
15
Epidemiological regions
16
We identified two main geographic regions that share similar HFMD epidemiological
17
characteristics: a northern region (latitude range 35·5°–46·2°N, plus Tibet, n=14) and
18
a southern region (19·5°–34·8°N, n=17) (Figure 7). Serotype–specific analysis
19
revealed a broadly similar split between Northern and Southern China, with a third
20
epidemiological region including Western provinces identified for EV-A71 and
14
1
CA-V16 (Appendix Figures 9–11). Discriminant analysis confirmed that climate
2
factors were the dominant predictors of HFMD epidemiological regions (Figure 7C,
3
Appendix Figures 9–11).
4
5
Discussion
6
Our study relies on a large sample of more than 7·2 million HFMD cases reported to
7
the national enhanced surveillance system during 2008–2012 in China and gives the
8
first comprehensive account of the national burden and epidemiology of the disease
9
(panel). The estimated HFMD incidence is 1·2 per 1,000 person–years in China and
10
the disease is responsible for 350–900 reported deaths annually, predominantly among
11
young children. As in other countries, we observed a predominance of EV-A71
12
serotype among severe cases, with a particularly young median age at infection of 2·3
13
years. Our large study covers 31 climatologically diverse provinces and suggests that
14
while HFMD cases tend to occur in warmer months of the year throughout the country,
15
peak timing and periodicity of epidemics varies with latitude. All serotypes causing
16
HFMD appear to follow broadly similar geographic gradients, which are moderately
17
associated with climate differences.
18
The observed age profile of infection is in agreement with reports from other
19
countries,4,20–22 and the particularly young mean age at infection suggests that
20
enteroviruses causing HFMD are highly transmissible. Relative sparing of infants
15
1
younger than six months was most likely due to protection by maternal antibodies.23,24
2
Infection with the causative viruses appears to confer lifelong immunity at least to
3
disease and possibly infection given the virtual absence of adult cases.
4
Our large-scale analysis of HFMD seasonal patterns identified two main
5
epidemiological regions corresponding to Northern China, where HFMD peaks in
6
summer, and Southern China, which experiences two HFMD peaks in spring and
7
autumn. A highly transmissible, strongly immunizing disease requires constant
8
replenishment of susceptibles to sustain semi-annual, even annual, epidemics.
9
Experience from Asia–Pacific countries has shown biennial25 or triennial20,21
10
outbreaks against a background total fertility rate of 1·3–3 births per woman,26
11
compared to China’s 1·6 births per woman under the longstanding one–child policy.27
12
In contrast, annual epidemics were reported in Japan and Malaysia,6,25 consistent with
13
observed patterns in Northern China, while semi-annual epidemics were reported in
14
Hong Kong SAR, South Taiwan of China, and Vietnam,8,10,21, in line with our
15
Southern China data. Further, our data indicate that EV-A71 epidemics occur annually
16
in China, in contrast to Japan and Malaysia where this serotype circulates every three
17
or four years.6,25 HFMD periodicity may be complicated by interference between the
18
causative enterovirus serotypes, perhaps associated with cross–serotype immunity.
19
Overall, the drivers of HFMD periodicity are not fully understood and worthy of
20
further study.
16
1
Although HFMD seasonal patterns were associated with precipitations, sunshine,
2
temperature, and air pressure, no single climatic variable explained the complexity of
3
HFMD seasonality across China. Our results confirm those of previous time series
4
analyses suggesting a relationship between HFMD and climatic factors in Singapore,
5
Hong-Kong, Southern China, and Japan.29–32 Moreover, the occurrence of
6
poliomyelitis, a disease caused by another enterovirus, may be associated with
7
humidity.33 Humidity was a weak predictor of HFMD seasonal patterns in our large
8
Chinese study, which encompasses a greater diversity of climate zones and
9
populations than previously studied. The HFMD epidemiological patterns in Hainan
10
and Tibet were apparently different from those in the other provinces (Fig. 4). This
11
divergence could be due to the extreme, unique climate of these two outliers: Hainan
12
is the only tropical region of China; while the climate in Tibet is very particular and
13
complex because of the great altitude. We cannot rule out artefactual surveillance
14
biases, in part due to the lack of financial and human resources (in Tibet in particular).
15
Clearly, more work is needed to clarify the local drivers of HFMD seasonality in
16
South East Asia.
17
Systematic, population–representative seroepidemiology studies across the age
18
spectrum from mother–infant pairs through the first ten years of life would greatly
19
refine our understanding of HFMD dynamics in different regions. Previous
20
cross–sectional serosurveys have provided age-specific estimates of the cumulative
17
1
incidence of infection at certain ages.34–37 Prospective longitudinal serological studies
2
could provide more detailed information on age–specific incidence of infection and
3
disease, risk factors for infection, and the protection against new infections conferred
4
by previous infections with the same or different viruses. Longitudinal serosurveys
5
were instrumental to elucidate the acquisition of infection and immunity in other
6
complex disease systems involving competition between viruses, which in turn
7
informed vaccination strategies.38
8
EV-A71 infection in early infancy conferred the gravest risks for clinical severity and
9
fatality.21,39 Several candidate vaccines against EV-A71 are currently undergoing
10
regulatory vetting in China.40 Whether and how these should be deployed require
11
further assessment of cost–effectiveness, feasibility and contextual considerations in
12
different regions. Transmission models are needed to fully assess the direct and
13
indirect benefits of vaccination, and the risk of serotype replacement given use of a
14
monovalent vaccine.
15
Our severity data highlight two independent risk factors: 1) each extra day since
16
symptom onset was associated with a 1% higher risk of mortality, and 2) those living
17
in a rural area had a one–quarter excess risk of death from infection. In addition to
18
rural–urban disparity in health care access and advanced life-sustaining technology,41
19
widespread and inappropriate use of glucocorticoids and pyrazolones in mild HFMD
20
stages42,43 could have precipitated critical illness and death.
18
1
Several limitations bear mention. First, as for any common, self-limiting illness,
2
surveillance only captures the tip of the clinical iceberg while most cases go
3
undetected because their condition is asymptomatic, or the patient does not seek
4
formal care, or he/she is not diagnosed and reported. Second, access to and provision
5
of health care as well as technical capacity varied between and sometimes within
6
provinces and there was no formal quality assurance or systematic audit for HFMD
7
surveillance. In particular, we cannot rule out that surveillance bias explained the
8
marked differences in HFMD patterns observed in Hainan and Tibet, relative to other
9
provinces (Fig. 4). Third, data collected during the first year of rollout are likely less
10
reliable than in more recent years. Fourth, we did not have data on mixed enterovirus
11
infections44 and were unable to serotype enteroviruses beyond CV-A16 and EV-A71.
12
In particular, we were unable to monitor serotype CV-A6, which has become more
13
predominant in Southern China.45 Fifth, some of severe neurological illnesses caused
14
by EV-A71 may have been missed by using a strict HFMD case definition. Sixth, our
15
study was based on a descriptive analysis of surveillance data. We unfortunately
16
cannot yet determine/estimate a threshold for yearly epidemic –this is beyond the
17
scope of our study.
18
19
In conclusion, we found substantial HFMD illness and mortality associated with
20
co-circulating EV-A71, CA-V16, and other enterovirus in China, disproportionately
19
1
affecting young children aged <5 years. Younger age, infection with EV-A71, and
2
living in rural area are risk factors for severe disease. Our study also uncovered
3
intriguing differences in HFMD seasonality across China, which may be associated
4
with climate. Overall, our study provides robust, population-based, national data to
5
prioritize EV-A71 vaccination and optimize the timing of routine vaccination
6
regionally. Further, our results will serve as a pre-vaccination baseline against which
7
future interventions can be compared. More broadly, further studies are warranted to
8
elucidate the dynamics and immunity patterns of HFMD, a rising public health
9
problem in Asia.
10
20
1
References
2
1.
Melnick JK. Enteroviruses: polioviruses, coxsackieviruses, echoviruses, and
3
newer enteroviruses. In: Fields BN, Knipe DM, Howley PM, Chanlock RM,
4
Melnick JL, Monath TP, et al., editors. Field's virology. 3rd ed. Philadelphia:
5
Lippincott-Raven; 1996. p. 655-712.
6
2.
Ooi MH, Wong SC, Lewthwaite P, Cardosa MJ, Solomon T. Clinical features,
7
diagnosis, and management of enterovirus 71. Lancet Neurol. 2010; 9:
8
1097-105.
9
3.
Schmidt NJ, Lennette EH, Ho HH. An apparently new enterovirus isolated from
10
patients with disease of the central nervous system. J Infect Dis. 1974; 129:
11
304-9.
12
4.
Chan LG, Parashar UD, Lye MS, Ong FG, Zaki SR, Alexander JP, et al. Deaths
13
of children during an outbreak of hand, foot, and mouth disease in sarawak,
14
malaysia: clinical and pathological characteristics of the disease. For the
15
Outbreak Study Group. Clin Infect Dis. 2000; 31: 678-83.
16
5.
Ho M, Chen ER, Hsu KH, Twu SJ, Chen KT, Tsai SF, et al. An epidemic of
17
enterovirus 71 infection in Taiwan. Taiwan Enterovirus Epidemic Working
18
Group. N Engl J Med. 1999; 341: 929-35.
19
20
6.
Hand, foot and mouth disease in Japan, 2002-2011. IASR. 33(3). Available at:
http://idsc.nih.go.jp/iasr/33/385/tpc385.html.
21
1
7.
Chan KP, Goh KT, Chong CY, Teo ES, Lau G, Ling AE. Epidemic hand, foot
2
and mouth disease caused by human enterovirus 71, Singapore. Emerg Infect
3
Dis. 2003; 9: 78-85.
4
8.
Tu PV, Thao NT, Perera D, Huu TK, Tien NT, Thuong TC, et al. Epidemiologic
5
and virologic investigation of hand, foot, and mouth disease, southern Vietnam,
6
2005. Emerg Infect Dis. 2007; 13: 1733-41.
7
9.
Zhang Y, Zhu Z, Yang W, Ren J, Tan X, Wang Y, et al. An emerging
8
recombinant human enterovirus 71 responsible for the 2008 outbreak of hand
9
foot and mouth disease in Fuyang city of China. Virol J. 2010; 7: 94.
10
10.
Ma E, Chan KC, Cheng P, Wong C, Chuang SK. The enterovirus 71 epidemic in
11
2008--public health implications for Hong Kong. Int J Infect Dis. 2010; 14:
12
e775-80.
13
11.
Seiff A. Cambodia unravels cause of mystery illness. Lancet. 2012; 380: 206.
14
12.
China Ministry of Health. Guideline for HFMD public health response. 2009.
15
Available at:
16
http://www.chinacdc.cn/jkzt/crb/szkb/jszl_2275/200906/t20090612_24707.html.
17
Accessed September 23 2013.
18
13.
Protocol of sample collection and laboratory tests for HFMD cases. 2009.
19
Available at:
20
http://www.chinacdc.cn/jkzt/crb/szkb/jszl_2275/200906/W02013010652285546
22
1
2
5929.pdf. Accessed October 25 2013.
14.
Zhang Y, Wang J, Guo W, Wang H, Zhu S, Wang D, et al. Emergence and
3
transmission pathways of rapidly evolving evolutionary branch C4a strains of
4
human enterovirus 71 in the Central Plain of China. PLoS One. 2011; 6: e27895.
5
15.
National Bureau of Statistics of China. National census in China in 2010.
6
Available at: http://www.stats.gov.cn/tjsj/pcsj/rkpc/6rp/indexch.html. Accessed
7
23 September 2012.
8
16.
9
10
http://www.zgjtnj.com/about/?2.html. Accessed June 15 2012.
17.
11
12
Yearbook of China integrated transport. Available at:
Climatic Data Center, National Meteorological Information Center, CMA.
Available at: http://data.cma.gov.cn/index.jsp. Accessed April 04 2013.
18.
Yu H, Alonso WJ, Feng L, Tan Y, Shu Y, Yang W, Viboud C. Characterization of
13
regional influenza seasonality patterns in China and implications for vaccination
14
strategies: spatio-temporal modelling of surveillance data. PLoS Med. 2013:
15
Nov 19. In Press.
16
19.
Naumova EN, Jagai JS, Matyas B, DeMaria A, Jr., MacNeill IB, Griffiths JK.
17
Seasonality in six enterically transmitted diseases and ambient temperature.
18
Epidemiol Infect. 2007; 135: 281-92.
19
20
20.
Ang LW, Koh BK, Chan KP, Chua LT, James L, Goh KT. Epidemiology and
control of hand, foot and mouth disease in Singapore, 2001-2007. Ann Acad
23
1
2
Med Singapore. 2009; 38: 106-12.
21.
Chen SC, Chang HL, Yan TR, Cheng YT, Chen KT. An eight-year study of
3
epidemiologic features of enterovirus 71 infection in Taiwan. Am J Trop Med
4
Hyg. 2007; 77: 188-91.
5
22.
6
7
Momoki ST. Surveillance of enterovirus infections in Yokohama city from 2004
to 2008. Jpn J Infect Dis. 2009; 62: 471-3.
23.
Wang SM, Ho TS, Lin HC, Lei HY, Wang JR, Liu CC. Reemerging of
8
enterovirus 71 in Taiwan: the age impact on disease severity. Eur J Clin
9
Microbiol Infect Dis. 2012; 31: 1219-24.
10
24.
Zhu FC, Liang ZL, Meng FY, Zeng Y, Mao QY, Chu K, et al. Retrospective
11
study of the incidence of HFMD and seroepidemiology of antibodies against
12
EV71 and CoxA16 in prenatal women and their infants. PLoS One. 2012; 7:
13
e37206.
14
25.
Podin Y, Gias EL, Ong F, Leong YW, Yee SF, Yusof MA, et al. Sentinel
15
surveillance for human enterovirus 71 in Sarawak, Malaysia: lessons from the
16
first 7 years. BMC Public Health. 2006; 6: 180.
17
26. The World Bank. Fertility rate, total (births per woman). Available at:
18
http://data.worldbank.org/indicator/SP.DYN.TFRT.IN?page=2. Accessed
19
February 20 2013.
20
27.
Hesketh T, Lu L, Xing ZW. The effect of China's one-child family policy after
24
1
2
25 years. N Engl J Med. 2005; 353: 1171-6.
28.
Ma E, Lam T, Chan KC, Wong C, Chuang SK. Changing epidemiology of hand,
3
foot, and mouth disease in Hong Kong, 2001-2009. Jpn J Infect Dis. 2010; 63:
4
422-6.
5
29.
6
7
disease. PLoS One. 2011; 6: e16796.
30.
8
9
Hii YL, Rocklov J, Ng N. Short term effects of weather on hand, foot and mouth
Ma E, Lam T, Wong C, Chuang SK. Is hand, foot and mouth disease associated
with meteorological parameters? Epidemiol Infect. 2010; 138: 1779-88.
31.
Onozuka D, Hashizume M. The influence of temperature and humidity on the
10
incidence of hand, foot, and mouth disease in Japan. Sci Total Environ. 2011;
11
410-411: 119-25.
12
32.
Huang Y, Deng T, Yu S, Gu J, Huang C, Xiao G, et al. Effect of meteorological
13
variables on the incidence of hand, foot, and mouth disease in children: a
14
time-series analysis in Guangzhou, China. BMC Infect Dis. 2013; 13: 134.
15
33.
16
17
Nathanson N, Kew OM. From emergence to eradication: the epidemiology of
poliomyelitis deconstructed. Am J Epidemiol. 2010; 172(11): 1213-29.
34.
Zhu Z, Zhu S, Guo X, Wang J, Wang D, Yan D, et al. Retrospective
18
seroepidemiology indicated that human enterovirus 71 and coxsackievirus A16
19
circulated wildly in central and southern China before large-scale outbreaks
20
from 2008. Virol J. 2010 Nov 4;7:300.
25
1
35.
Zhu FC, Liang ZL, Meng FY, Zeng Y, Mao QY, Chu K, et al. Retrospective
2
study of the incidence of HFMD and seroepidemiology of antibodies against
3
EV71 and CoxA16 in prenatal women and their infants. PLoS One.
4
2012;7(5):e37206.
5
36.
Zeng M, El Khatib NF, Tu S, Ren P, Xu S, Zhu Q, et al. Seroepidemiology of
6
Enterovirus 71 infection prior to the 2011 season in children in Shanghai. J Clin
7
Virol. 2012 Apr;53(4):285-9.
8
37.
9
against enterovirus 71 in children from Lu'an City in Central China. Jpn J Infect
10
11
Yu H, Wang M, Chang H, Lu J, Lu B, Li J, et al. Prevalence of antibodies
Dis. 2011;64(6):528-32.
38.
Velazquez FR, Matson DO, Calva JJ, Guerrero L, Morrow AL, Carter-Campbell
12
S, et al. Rotavirus infections in infants as protection against subsequent
13
infections. N Engl J Med. 1996; 335: 1022-8.
14
39.
Chang LY, King CC, Hsu KH, Ning HC, Tsao KC, Li CC, et al. Risk factors of
15
enterovirus 71 infection and associated hand, foot, and mouth
16
disease/herpangina in children during an epidemic in Taiwan. Pediatrics. 2002;
17
109: e88.
18
40.
19
20
Liang Z, Mao Q, Gao F, Wang J. Progress on the research and development of
human enterovirus 71 (EV71) vaccines. Front Med. 2013; 7: 111-21.
41.
Shi L. Health care in China: a rural-urban comparison after the socioeconomic
26
1
2
reforms. Bull World Health Organ. 1993; 71: 723-36.
42.
Jiang Q, Yu BN, Ying G, Liao J, Gan H, Blanchard J, et al. Outpatient
3
prescription practices in rural township health centers in Sichuan Province,
4
China. BMC Health Serv Res. 2012; 12: 324.
5
43.
Ma H, He F, Wan J, Jin D, Zhu L, Liu X, et al. Glucocorticoid and pyrazolone
6
treatment of acute fever is a risk factor for critical and life-threatening human
7
enterovirus 71 infection during an outbreak in China, 2008. Pediatr Infect Dis J.
8
2010; 29: 524-9.
9
44.
Yan XF, Gao S, Xia JF, Ye R, Yu H, Long JE. Epidemic characteristics of hand,
10
foot, and mouth disease in Shanghai from 2009 to 2010: Enterovirus 71
11
subgenotype C4 as the primary causative agent and a high incidence of mixed
12
infections with coxsackievirus A16. Scand J Infect Dis. 2012; 44(4): 297-305.
13
45.
He YQ, Chen L, Xu WB, Yang H, Wang HZ, Zong WP, et al. Emergence,
14
circulation, and spatiotemporal phylogenetic analysis of coxsackievirus a6- and
15
coxsackievirus a10-associated hand, foot, and mouth disease infections from
16
2008 to 2012 in Shenzhen, China. J Clin Microbiol. 2013; 51(11): 3560-6.
27
Table 1. Estimated rates of incidence, severe illness and mortality amongst all HFMD cases by age group
Incidence rates (95% confidence interval),
Age
per 1,000,000
Severe illness rates (95% confidence
Mortality rates (95% confidence
interval), per 1,000,000
interval), per 1,000,000
group
Total
2008
2009
2010
2011
2012
2008
2009
2010
2011
2012
2008
2009
2010
2011
2012
369.2
869.3
1352.9
1221.3
1616.4
0.9
10.4
21.3
14.1
15.6
0.1
0.3
0.7
0.4
0.4
(368.2-
(867.8-
(1350.9-
(1219.5-
(1614.3-
(0.9-
(10.2-
(21.0-
(13.9-
(15.4-
(0.1-
(0.2-
(0.6-
(0.4-
(0.4-
370.3)
870.9)
1354.9)
1223.2)
1618.6)
1.0)
10.6)
21.5)
14.3)
15.8)
0.1)
0.3)
0.7)
0.4)
0.5)
673.7
1319.9
2259.6
2372.5
4014.7
3.8
35.9
60.5
45.5
65.0
0.6
2.0
3.3
2.4
1.5
(655.5-
(1294.9-
(2227.1-
(2338.0-
(3966.9-
(2.6-
(31.7-
(55.1-
(40.7-
(58.9-
(0.2-
(1.2-
(2.2-
(1.4-
(0.7-
691.8)
1345.0)
2292.1)
2407.1)
4062.5)
5.5)
40.0)
65.8)
50.2)
71.1)
1.5)
3.2)
4.8)
3.7)
2.7)
7186.5
18561.8
28042.1
25539.2
28862.1
27.3
317.1
589.6
336.9
407.4
3.8
10.5
27.1
11.5
13.3
(7132.1-
(18475.6-
(27937.1-
(25435.2-
(28744.5-
(23.9
(305.9-
(574.4-
(325.0-
(393.4-
(2.6-
(8.5-
(23.9-
7241.0)
18647.9)
28147.1)
25643.2)
28979.7)
-30.6)
328.4)
604.8)
348.9)
421.4)
5.2)
12.6)
30.4)
13.7)
15.9)
8180.8
19148.5
29480.4
30215.9
38191.1
26.4
326.8
626.1
457.3
492.4
3.2
8.6
20.2
12.9
15.3
(8137.1-
(19082.5-
(29399.1-
(30130.6-
(38096.4-
(23.9-
(318.2-
(614.3-
(446.8-
(481.6-
(2.3-
(7.2-
(18.1-
(11.2-
(13.4-
8224.4)
19214.4)
29561.6)
30301.2)
38285.8)
28.9)
335.4)
637.9)
467.8)
503.1)
4.1)
10.0)
22.4)
14.7)
17.2)
4760.6
11306.9
16848.0
16815.9
23111.3
8.8
93.2
204.8
163.3
182.1
0.6
1.8
5.2
3.9
4.1
(4741.1-
(11276.9-
(16811.4-
(16777.9-
(23067.3-
(8.0-
(90.5-
(200.8-
(159.5-
(178.2-
(0.4-
(1.4-
(4.6-
4780.1)
11336.8)
16884.5)
16853.8)
23155.3)
9.6)
95.9)
208.8)
167.0)
186.0)
0.9)
2.2)
5.9)
<6
months
6-11
(9.3-
(10.8-
months
12-23
months
24-59
(3.3-
(3.5-
months
4.5)
4.6)
28
Table 1. (continued)
Incidence rates (95% confidence interval),
Age
per 1,000,000
Severe illness rates (95% confidence
Mortality rates (95% confidence
interval), per 1,000,000
interval), per 1,000,000
group
2008
2009
2010
2011
2012
2008
2009
2010
2011
2012
2008
2009
2010
2011
2012
580.7
1045.1
1712.0
1565.3
2706.5
0.7
5.2
10.4
8.8
10.2
0.0
0.1
0.1
0.1
0.1
(575.3-
(1037.8-
(1702.8-
(1556.2-
(2694.5-
(0.5-
(4.7-
(9.7-
(8.1-
(9.5-
(0.0-
(0.0-
(0.1-
(0.0-
(0.1-
586.2)
1052.4)
1721.2)
1574.4)
2718.5)
0.9)
5.7)
11.1)
9.5)
10.9)
0.1)
0.1)
0.3)
0.2)
0.2)
68.4
132.3
216.3
186.7
277.2
0.1
0.3
1.1
0.9
0.8
0.0
0.0
0.0
0.0
0.0
(66.7-
(129.9-
(213.0-
(183.5-
(273.4-
(0.0-
(0.2-
(0.9-
(0.7-
(0.0-
(0.0-
(0.0-
(0.0-
(0.0-
70.1)
134.7)
219.5)
189.9)
281.0)
0.1)
0.5)
1.4)
1.2)
1.0)
0.0)
0.0)
0.1)
0.1)
0.1)
≥15
2.3
4.1
6.5
5.9
7.7
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
(0.0-
(0.0-
(0.0-
(0.0-0.0
(0.0-0.
(0.0-
(0.0-
(0.0-
(0.0-
(0.0-
years
(2.2-2.4)
(4.0-4.2)
(6.4-6.7)
(5.8-6.0)
(7.6-7.9)
0.0)
0.0)
0.0)
)
0)
0.0)
0.0)
0.0)
0.0)
0.0)
5-9
years
10-14
years
(0.6-
29
Table 2. ORs and 95%CIs of severe illness and death among probable cases and
laboratory-confirmed HFMD cases. Results from multivariate logistic regression
with backward selection of demographic and geographic variables, time from onset to
diagnosis, and virus type (p-value for exclusion≥0·10).
Lab-confirmed cases
Probable cases
N=267942
N=6932150
No. of
Adjusted OR (95%
No. of
Adjusted OR (95%
cases
CI)
cases
CI)
CV-A16
76817
Ref.
NA
Other enterovirus
56398
4.36 (2.72-6.98)
EV-A71
134727
34.24 (22.47-52.18)
≥5 years
22666
Ref.
638925
Ref.
24-59 months
131560
3.44 (2.24-5.28)
3274253
4.00 (2.12-7.55)
6-23 months
111238
8.75 (5.73-13.35)
2941707
8.59 (4.59-16.09)
<6 months
2478
20.46 (12.35-33.90)
77265
14.21 (6.59-30.62)
Urban
115828
Ref.
3219524
Ref.
Rural
137162
1.19 (1.07-1.33)
3218311
2.45 (2.06-2.92)
NA
1.01 (1.01-1.02)
NA
1.02 (1.01-1.02)
Female
96934
Ref.
4351763
Ref.
Male
171008
1.00 (0.90-1.11)
2580384
1.28 (1.09-1.52)
CV-A16
76817
Ref.
NA
Other enterovirus
56398
3.91 (3.70-4.14)
EV-A71
134727
11.17 (10.64-11.74)
22666
Ref.
638925
Ref.
24-59 months
131560
1.90 (1.78-2.01)
3274253
1.99 (1.89-2.10)
6-23 months
111238
3.64 (3.43-3.87)
2941707
3.40 (3.22-3.58)
<6 months
2478
5.68 (5.07-6.36)
77265
4.71 (4.33-5.12)
Characteristics
Death
Enterovirus serotype
Age group
Residence
Per day increase of
onset-to-diagnosis
interval
Sex
Severe illness
Enterovirus serotype
Age group
≥5 years
CI, confidence interval; EV-A71, enterovirus 71; CA-V16, Coxsackievirus A16; Ref.,
reference; NA, not applicable; OR, odds ratio
30
Table 2. (continued)
Lab-confirmed cases
Probable cases
N=267942
N=6932150
No. of
Adjusted OR (95%
No. of
Adjusted OR (95%
cases
CI)
cases
CI)
Urban
115828
Ref.
3219524
Ref.
Rural
137162
1.08 (1.05-1.10)
3218311
1.67 (1.64-1.71)
NA
1.00 (1.00-1.00)
NA
1.02 (1.01-1.02)
Characteristics
Severe illness
Residence
Per day increase of
onset-to-diagnosis
interval
Sex
Female
96934
Male
171008
4351763
1.10 (1.07-1.13)
2580384
1.03 (1.01-1.05)
CI, confidence interval; EV-A71, enterovirus 71; CA-V16, Coxsackievirus A16; Ref.,
reference; NA, not applicable; OR, odds ratio
31
Panel
Systematic review
We searched PubMed with the keywords “Hand, Foot and Mouth Disease”, “EV71”,
“enterovirus 71” and “coxsackie virus A” for all journal articles written in English
between Jan 1, 1995, and June 30, 2013. In the past 15 years, EV-A71–associated
HFMD epidemics have been increasingly reported across the Asia–Pacific region,
chronologically including Malaysia,4 Taiwan,5 Japan,6 Singapore,7 Vietnam,8
mainland China,9 Hong Kong SAR,10 and Cambodia.11 These reports were mostly
descriptive, with very few exceptions that estimated epidemiologic parameters,
assessed severity or analyzed seasonality. Even fewer had sufficient power to yield
definitive conclusions. No previous report had addressed all three sets of interrelated
questions. Here we report the largest study to date across a wide geographic expanse
through the recently enhanced national surveillance network for HFMD in China.
Interpretation
We described the enhanced HFMD surveillance system, characterized host
characteristics, periodicity of epidemics, and geographic patterns from 2008 to 2012.
HFMD incidence was 1·2 per 1,000 person–years with 350–900 reported deaths
annually during 2009–2012, predominantly among young children. There was strong
age dependency, with children aged six months to two years showing the highest rates
(Table 1). Boys under five years were 1·6 times more likely than girls of the same age
32
bracket to register disease. Overall the case–fatality, case–severity, and severe
case–fatality risks were 0·03%, 1·1%, and 3·0% respectively. Illness severity and
death by enterovirus serotype were inversely related to age (Table 2 and Figure 2,
panels B–D). The median time from illness onset to diagnosis, from illness onset to
death, and from diagnosis to death were 1·5 days (interquartile range: 0·5–2·5 days),
3·5 days (interquartile range: 2·5–4·5 days), and 0·5 day (interquartile range: 0·5–1·5
days) respectively. HFMD tended to occur during the warmer months of the year
throughout the country, although peak timing and the periodicity of epidemics varied
substantially with latitude, in particular demonstrating stronger semi-annual cycles in
southern China compared to annual outbreaks in the north. Seasonal variation was
associated with precipitation, sunshine, temperature, and barometric pressure.
However, no single climatic variable was sufficient to explain the complexity of
HFMD seasonality across China.
33
Figure Legends
Figure 1. Proportions of enterovirus serotypes among laboratory-confirmed
HFMD cases by clinical severity, 2008-2012, China. A: based on mild cases. B:
based on severe cases who survived, C: based on fatal cases.
Figure 2. Age distribution and clinical severity of probable and
laboratory-confirmed (EV-A71, CA-V16 or other enteroviruses) HFMD cases,
2008-2012, China. A: Age distribution of probable and lab-confirmed cases. B: Risk
of fatality among cases by age group and viral etiology. C: Risk of severe illness
among cases by age group and viral etiology. D: Risk of fatality among severe cases
by age group and viral etiology. EV-A71, enterovirus 71; CA-V16, Coxsackievirus
A16; mth, months; y, years. Severity estimates in B-D were calculated by
extrapolating the serotype distribution among test-positive cases to untested and
test-negative cases. That is, the number of mild cases with serotype X ( = EV-A71,
CA-V16 or other enteroviruses) was estimated to be (no. of mild cases test-positive
for serotype X)/(no. of test-positive mild cases) x (no. of mild cases); severe cases
were similarly analyzed. Results were similar if only the 2010-2012 or 2012 data were
used (Appendix Figures 4&5).
Figure 3. Estimates of onset-to-diagnosis, onset-to-death, and diagnosis-to-death
34
distributions of probable and laboratory-confirmed (EV-A71, CA-V16 or other
enteroviruses) HFMD cases, 2008-2012, China. A: Onset-to-diagnosis distribution
by viral etiology (n=7,200,092). B: Onset-to-death distribution by viral etiology
(n=2457). C: Diagnosis-to-death distribution by viral etiology (n=2457). Note that
some intervals are negative because diagnosis occurred after death.
Figure 4. Heatmap of HFMD surveillance data from 2008 to 2012 by Chinese
province. The provinces were ordered by latitude from Northermost (top) to
Southernmost (bottom). A: Time series of weekly probable and lab-confirmed HFMD
cases, standardized by the number of annual cases. B: Seasonal distribution of HFMD
cases, plotted as the median value of proportion of cases in each week of the year
from 2008 to 2012. C: Number of HFMD cases by week of illness onset. The insert is
a superposition of the number of cases without probable HFMD cases by week of
illness onset.
Figure 5. Latitudinal gradients in periodicity and peak timing of HFMD. A:
Amplitude of the annual periodicity. B: Annual peak timing. C: Contribution of the
semi-annual periodicity, measured by the ratio of the amplitude of the semi-annual
periodicity to the sum of the amplitudes of annual and semi-annual periodicities
(higher ratio indicates a stronger semi-annual periodicity). Symbol size is proportional
to the number of cases in each province. Black solid line represent linear regression fit
35
(regression weighted by mean annual number of HFMD cases). P-values are given on
the graphs. Colors represent different climatic zones (black: cold-temperate, blue
mid-temperate, green warm-temperate, orange subtropical, red tropical).
Figure 6. Amplitude and timing of primary HFMD epidemics in China. A:
Amplitude of the annual cycle from yellow (low) to red (high), as indicated in the
legend. B: Importance of the semi-annual periodicity, measured by the ratio of the
amplitude of the semi-annual cycle to the sum of the amplitudes of annual and
semi-annual cycles. Pale green indicates strongly annual influenza epidemics, while
dark green indicates dominant semi-annual activity. C: Timing of primary annual
HFMD peak, in weeks from Jan 1st. Timing is color coded from pale blue to dark
blue.
Figure 7. HFMD epidemiological regions and predictors. A: Identified
epidemiological regions based on hierarchical clustering, using the Euclidian distance
between weekly standardized HFMD time series. Provinces are color-coded by
climatic region (black: cold-temperate, blue: mid-temperate, green: warm temperate,
orange: subtropical, red: tropical). B: Map of the three epidemiological regions
identified in panel A. C: Climate predictors of the two main clusters identified in
panel A, based on stepwise discriminant analysis.
36
Supporting Information
Appendix Text 1
Additional information regarding climate variables.
Appendix Text 2
Additional information regarding the estimation of seasonal
parameters.
Appendix Table 1
Association between HFMD, EV-A71, CA-V16 and other
enteroviruses seasonal patterns and geographic, population, economic, transport and
climatic factors.
Appendix Figure 1 Flowchart of data and sample collection for HFMD cases,
2008-2012, China.
Appendix Figure 2 Map of the Chinese mainland provinces.
Appendix Figure 3 Proportions of enterovirus serotypes among
laboratory-confirmed HFMD cases by week of illness onset and by geographic region,
2008-2012, China.
Appendix Figure 4 Age distribution and clinical severity of probable and
laboratory-confirmed HFMD cases, 2010-2012, China
Appendix Figure 5 Age distribution and clinical severity of probable and
laboratory-confirmed HFMD cases, 2012, China
Appendix Figure 6 Latitudinal and longitudinal gradients in seasonality of EV-A71,
CA-V16 and other enterovirus.
Appendix Figure 7 Periodicity of EV-A71, CA-V16 and other enterovirus
epidemics in China.
37
Appendix Figure 8 Importance of semi-annual seasonality of EV-A71, CA-V16 and
other enterovirus.
Appendix Figure 9 EV-A71 epidemiological regions and predictors.
Appendix Figure 10 CA-V16 epidemiological regions and predictors.
Appendix Figure 11 Other enteroviruses epidemiological regions and predictors.
Competing Interests
The authors have no competing interests to declare.
Acknowledgments
We thank the hospitals, local health departments and Centers for Disease Control and
Prevention in China for assistance in coordinating data collection. We also thank Lin
Wang for the data cleaning. We are also grateful to the National Health and Family
Planning Commission for supporting this study. The views expressed are those of the
authors and do not necessarily represent the policy of the China CDC.
Abbreviations
95% CI, 95% confidence interval; CA-V16, coxsackievirus A16; CDC, Centre for
Disease Control and Prevention; CPE, cytopathic effect; EV-A71, Enterovirus 71;
HFMD, Hand-Foot-And-Mouth Disease; RD, rhabdomyosarcoma; RNA, ribonucleic
acid; RT-PCR, reverse-transcriptase polymerase chain reaction
38
Figure 1. Proportions of enterovirus serotypes among laboratory-confirmed
HFMD cases by clinical severity, 2008-2012, China
A: based on mild cases. B: based on severe cases who survived, C: based on fatal
cases. EV-A71, enterovirus 71; CV-A16, Coxsackievirus A16
Figure 2. Age distribution and clinical severity of probable and laboratory-confirmed HFMD cases,
2008-2012, China
A: Age distribution of probable and lab-confirmed cases. B: Risk of fatality among cases by age group and viral
etiology. C: Risk of severe illness among cases by age group and viral etiology. D: Risk of fatality among
severe cases by age group and viral etiology. EV-A71, enterovirus 71; CV-A16, Coxsackievirus A16; mth,
months; y, years. Severity estimates in B-D were calculated by extrapolating the serotype distribution among
test-positive cases to untested and test-negative cases. That is, the number of mild cases with serotype X ( =
EV-A71, CV-A16 or other enteroviruses) was estimated to be (no. of mild cases test-positive for serotype
X)/(no. of test-positive mild cases) x (no. of mild cases); severe cases were similarly analyzed. Results were
similar if only the 2010-2012 or 2012 data were used (Appendix Figures 4&5).
Figure 3. Estimates of onset-to-diagnosis, onset-to-death, and diagnosis-to-death distributions of probable and laboratory-confirmed
HFMD cases, 2008-2012, China
A: Onset-to-diagnosis distribution by viral etiology (n=7,200,092). B: Onset-to-death distribution by viral etiology (n=2457). C:
Diagnosis-to-death distribution by viral etiology (n=2457). Note that some intervals are negative because diagnosis occurred after death. EV-A71,
enterovirus 71; CV-A16, Coxsackievirus A16
A
B
C
Figure 4. HFMD surveillance data from 2008 to 2012 by Chinese province and
by week
The provinces were ordered by latitude from Northermost (top) to Southernmost
(bottom). A: Time series of weekly probable and lab-confirmed HFMD cases,
standardized by the number of annual cases. B: Seasonal distribution of HFMD
cases, plotted as the median value of proportion of cases in each week of the year
from 2008 to 2012. C: Number of HFMD cases by week of illness onset. The insert
is a superposition of the number of cases without probable HFMD cases by week of
illness onset.
Figure 5. Latitudinal gradients in periodicity and peak timing of HFMD
A: Amplitude of the annual periodicity. B: Annual peak timing. C: Contribution of the
semi-annual periodicity, measured by the ratio of the amplitude of the semi-annual
periodicity to the sum of the amplitudes of annual and semi-annual periodicities
(higher ratio indicates a stronger semi-annual periodicity). Symbol size is proportional
to the number of cases in each province. Black solid line represent linear regression fit
(regression weighted by mean annual number of HFMD cases). P-values are given on
the graphs. Colors represent different climatic zones (black: cold-temperate, blue
mid-temperate, green warm-temperate, orange subtropical, red tropical).
Figure 6. Amplitude and timing of primary HFMD epidemics in China.
A: Amplitude of the annual cycle from yellow (low) to red (high), as indicated in the legend. B: Importance of the semi-annual periodicity,
measured by the ratio of the amplitude of the semi-annual cycle to the sum of the amplitudes of annual and semi-annual cycles. Pale green
indicates strongly annual influenza epidemics, while dark green indicates dominant semi-annual activity. C: Timing of primary annual HFMD
peak, in weeks from Jan 1st. Timing is color coded from pale blue to dark blue.
A
B
C
Figure 7. HFMD epidemiological regions and predictors
A: Identified epidemiological regions based on hierarchical clustering, using the Euclidian
distance between weekly standardized HFMD time series. Provinces are color-coded by climatic
region (black: cold-temperate, blue: mid-temperate, green: warm temperate, orange: subtropical,
red: tropical). B: Map of the three epidemiological regions identified in panel A. C: Climate
predictors of the two main clusters identified in panel A, based on stepwise discriminant analysis.
Appendix Text 1. Additional information regarding climate variables
Meteorological data collected included daily temperature (min, max, mean), rainfall,
relative humidity (min, mean), air pressure (min, max, mean), average vapor pressure,
and hours of sunshine recorded at 756 Chinese ground weather stations from
2008–2012.1
In order to aggregate daily- station-specific meteorological data at the weekly and
province level, we first averaged station–specific daily values at the weekly level
(except for rainfall where the weekly sum was taken) and took the average across all
stations located within a province. Preliminary analysis of HFMD surveillance data
revealed that two HFMD peaks were typically observed in recent years in China in
spring and autumn. Therefore, province–specific climatic variables were further
summarized for relevant seasons, by averaging the weekly values corresponding to
the spring (April–June) and autumn (September–November).
Reference
1.
Climatic Data Center, National Meteorological Information Center, CMA.
Available at: http://data.cma.gov.cn/index.jsp. Accessed April 04 2013.
1
Appendix Text 2. Additional information regarding the estimation of seasonal
parameters
1. Regression model
We used to following linear regression model to estimate the peak timing and
amplitude of the annual and semi-annual periodicities of influenza activity in each
province, as described in1,2:
hfmdi(t)=a+ bi *cos(2Pi*t/52.17) + ci *sin(2Pi*t/52.17) + di*cos(4Pi*t/52.17) +
ei *sin(4Pi*t/52.17) + ε(t)
where hfmdi(t) are the weekly standardized counts of HFMD cases (or EV-A71, or
CV-A16, or other enterovirus) isolates in province i, where standardization is
obtained by dividing weekly values by the annual number of HFMD cases, t is a
running index for week, and a,b, c, d and e are the intercept and seasonal terms to be
estimated from the data.
Specifically,
the amplitude of the annual periodicity is estimated as AnnAmpi=sqrt(bi 2+ ci 2),
the annual peak timing is estimated as AnnPeakTimingi=-atan(ci / bi),
the amplitude of the semi-annual periodicity is estimated as
SemiAnnAmpi=sqrt(c2+d2)
To evaluate the relative importance of annual and semi-annual HFMD periodicities by
province, we calculated the ratio between the amplitude of the semi-annual periodicity
and the sum of the amplitudes of annual and semi-annual periodicities. A ratio close to
1 is indicative of dominant semi-annual periodicity while a ratio close to 0 indicates
2
dominant annual periodicity.
The ratio is estimated as Ratioi= SemiAnnAmpi / (SemiAnnAmpi + AnnAmpi)
To control for different levels of HFMD activity across provinces, we compare the
relative amplitudes of annual and semi-annual periodicity, obtained by dividing
AnnAmpi and SemiAnnAmpi by the mean of the hfmdi(t) time series.
2. Sensitivity analyses
We considered alternative seasonal regression models using a Poisson distribution
and a log link (given that we were using modified disease counts), with or without an
offset for the number of specimens tested. Given that all three approaches gave
sensibly similar results, we elected to use the linear model in the main analysis for
the sake of simplicity.
Reference
1.
Naumova EN, Jagai JS, Matyas B, DeMaria A, Jr., MacNeill IB, Griffiths JK.
Seasonality in six enterically transmitted diseases and ambient temperature.
Epidemiol Infect. 2007; 135: 281-92.
2.
Yu H, Alonso WJ, Feng L, Tan Y, Shu Y, Yang W, Viboud C. Characterization of
regional influenza seasonality patterns in China and implications for vaccination
strategies: spatio-temporal modelling of surveillance data. PLoS Med. 2013:
Nov 19. In Press
3
Appendix Table 1. Association between HFMD, EV71, CA16 and other enteroviruses seasonal patterns and geographic, population,
economic, transport and climatic factors
Analyzed
Overall cases
EV71 confirmed cases
Timing (phase)
Amplitude
Importance
Timing (phase)
Amplitude
Importance
factors
Latitude
Annual
cycle
(p-value,
R2)
<0.001,
0.12
Longitude
0.032,
0.03
Population size
<0.001,
0.19
Area
0.021,
0.04
Density
0.032,
0.03
Semi-annual
cycle
(p-value, R2)
Semi-annual
cycle
(p-value, R2)
semi-annual
periodicity
(p-value, R2)
Annual
cycle
(p-value
, R2)
Semi-annual
cycle
(p-value, R2)
Annual
cycle
(p-value,
R2)
Semi-annual
cycle
(p-value, R2)
0.001,
0.05
0.016,
0.09
0.012,
0.06
<0.001,
0.09
0.006,
0.03
by waterways
0.006,
0.03
by airways
0.008,
0.03
0.004,
0.20
<0.001,
0.06
0.004,
0.03
0.006,
0.03
0.018,
0.04
semi-annual
periodicity
(p-value, R2)
0.002,
0.15
0.048,
0.05
Passengers
by railways
by highways
Annual
cycle
(p-value,
R2)
<0.001,
0.07
0.013,
0.11
0.005,
0.21
0.025,
0.11
0.005,
0.21
0.025,
0.11
0.005,
0.21
0.025,
0.11
0.038,
0.01
Analyses were based on stepwise multivariate regression
1
Appendix Table 1. (continued)
Analyzed
Timing (phase)
factors
Annual
cycle
(p-value,
R2)
Total
passengers
GRP per capita 0.029,
0.03
Rainfall1
Sunshine1
Air pressure1
max
<0.001,
0.12
Overall cases
EV71 confirmed cases
Semi-annual
cycle
(p-value, R2)
Amplitude
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
Importance
semi-annual
periodicity
(p-value, R2)
0.019,
0.04
0.006,
0.03
0.037,
0.04
Timing (phase)
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
<0.001,
0.08
0.001,
0.06
0.003,
0.05
<0.001,
0.08
0.019,
0.04
0.005,
0.21
0.025,
0.11
0.038,
0.10
0.027,
0.11
0.016,
0.08
0.023,
0.02
min
0.009,
0.03
0.014,
0.08
0.04,
0.02
<0.001,
0.23
min
Importance
semi-annual
periodicity
(p-value, R2)
0.031,
0.08
0.018,
0.10
mean
Relative
humidity1
mean
Amplitude
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
<0.001,
0.26
1
Spring weather parameters
2
Appendix Table 1. (continued)
Analyzed
Timing (phase)
factors
Annual
cycle
(p-value,
R2)
Temperature1
max
Semi-annual
cycle
(p-value, R2)
Overall cases
Amplitude
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
0.036,
0.03
0.007,
0.03
min
0.007,
0.03
Rainfall2
Sunshine2
Air pressure2
max
mean
Timing (phase)
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
Amplitude
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
Importance
semi-annual
periodicity
(p-value, R2)
0.019,
0.07
mean
Vapor
pressure1
EV71 confirmed cases
Importance
semi-annual
periodicity
(p-value, R2)
0.049,
0.05
0.022,
0.04
0.003,
0.08
0.006,
0.06
0.017,
0.09
0.045,
0.01
0.024,
0.04
0.009,
0.11
<0.001,
0.21
0.032,
0.07
<0.001,
0.08
<0.001,
0.11
0.032,
0.03
min
0.009,
0.17
1
Spring weather parameters
Autumn weather parameters
2
3
Appendix Table 1. (continued)
Analyzed
Timing (phase)
factors
Annual
cycle
(p-value,
R2)
Semi-annual
cycle
(p-value, R2)
Overall cases
Amplitude
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
Temperature2
max
EV71 confirmed cases
Importance
semi-annual
periodicity
(p-value, R2)
Timing (phase)
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
0.027,
0.07
Amplitude
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
Importance
semi-annual
periodicity
(p-value, R2)
<0.001,
0.06
mean
0.010,
0.11
0.043,
0.04
<0.001,
0.06
min
0.012,
0.10
0.033,
0.04
<0.001,
0.06
0.044,
0.10
0.048,
0.08
0.397
0.457
Relative
humidity2
mean
<0.001,
0.07
min
0.044,
0.03
Vapor
pressure2
Total explained
variance3
0.832
0.817
0.030,
0.05
0.910
0.623
0.755
0.671
0.915
0.646
2
Autumn weather parameters
Explained variance by multivariate best fit models selected by stepwise regression
3
4
Appendix Table 1. (continued)
Analyzed
CA16 confirmed cases
Timing (phase)
Amplitude
factors
Latitude
Longitude
Annual
cycle
(p-value,
R2)
<0.001,
0.16
Semi-annual
cycle
(p-value, R2)
<0.001,
0.23
Annual
cycle
(p-value,
R2)
0.011,
0.07
0.003,
0.10
Other enterovirus confirmed cases
Semi-annual
cycle
(p-value, R2)
Importance
semi-annual
periodicity
(p-value, R2)
<0.001,
0.41
0.009,
0.16
Timing (phase)
Annual
Semi-annual
cycle
cycle
(p-value,
(p-value,
R2)
R2)
0.007,
0.05
Area
0.028,
0.05
Density
0.011,
0.07
Passengers
by railways
by highways
by airways
Importance
semi-annual
periodicity
(p-value, R2)
0.002,
0.21
Population size
by waterways
Amplitude
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
0.004,
0.09
0.007,
0.09
0.040,
0.07
0.034,
0.06
0.027,
0.06
0.003,
0.13
0.003,
0.12
0.003,
0.13
0.003,
0.12
0.003,
0.13
0.002,
0.14
0.048,
0.06
0.021,
0.04
0.003,
0.12
0.042,
0.05
Analyses were based on stepwise multivariate regression
5
Appendix Table 1. (continued)
Analyzed
CA16 confirmed cases
Timing (phase)
Amplitude
factors
Total
passengers
Annual
cycle
(p-value,
R2)
Semi-annual
cycle
(p-value, R2)
0.018,
0.06
0.003,
0.13
Rainfall1
Sunshine1
mean
min
0.008,
0.09
0.012,
0.15
0.005,
0.06
<0.001,
0.23
0.034,
0.05
0.032,
0.05
Amplitude
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
Importance
semi-annual
periodicity
(p-value, R2)
0.006,
0.12
0.005,
0.12
0.012,
0.08
0.006,
0.12
0.010,
0.12
0.020,
0.05
0.008,
0.13
0.008,
0.08
min
Relative
humidity1
mean
Semi-annual
cycle
(p-value, R2)
Timing (phase)
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
0.003,
0.12
0.049,
0.03
<0.001,
0.23
GRP per capita
Air pressure1
max
Annual
cycle
(p-value,
R2)
Other enterovirus confirmed cases
Importance
semi-annual
periodicity
(p-value, R2)
0.004,
1.10
0.001,
0.13
0.003,
0.17
1
Spring weather parameters
6
Appendix Table 1. (continued)
Analyzed
CA16 confirmed cases
Timing (phase)
Amplitude
factors
Annual
cycle
(p-value,
R2)
Semi-annual
cycle
(p-value, R2)
Temperature1
max
Annual
cycle
(p-value,
R2)
Semi-annual
cycle
(p-value, R2)
0.020,
0.05
Other enterovirus confirmed cases
Importance
semi-annual
periodicity
(p-value, R2)
Timing (phase)
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
Amplitude
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
0.014,
0.10
mean
0.010,
0.07
0.025,
0.07
min
0.008,
0.08
0.025,
0.07
0.005,
0.14
Vapor
pressure1
0.005,
0.10
Rainfall2
Sunshine2
Importance
semi-annual
periodicity
(p-value, R2)
0.004,
0.12
Air pressure2
max
0.033,
0.05
mean
min
0.046,
0.05
<0.001,
0.11
0.003,
0.12
0.032,
0.09
0.005,
0.06
0.034,
0.05
0.22,
0.04
0.018,
0.04
<0.001,
0.27
0.007,
0.15
<0.001,
0.17
0.009,
0.14
1
Spring weather parameters
Autumn weather parameters
2
7
Appendix Table 1. (continued)
Analyzed
CA16 confirmed cases
Timing (phase)
Amplitude
factors
Annual
cycle
(p-value,
R2)
Semi-annual
cycle
(p-value, R2)
Annual
cycle
(p-value,
R2)
Temperature2
max
Timing (phase)
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
Amplitude
Annual
Semi-annual
cycle
cycle
(p-value, (p-value, R2)
R2)
0.002,
0.13
0.002,
0.13
mean
min
Relative
humidity2
mean
Importance
semi-annual
periodicity
(p-value, R2)
0.020,
0.11
0.011,
0.13
0.007,
0.09
0.007,
0.09
0.015,
0.07
0.014,
0.07
0.017,
0.07
min
0.001,
0.17
Vapor
pressure2
Total explained
variance3
Semi-annual
cycle
(p-value, R2)
Other enterovirus confirmed cases
Importance
semi-annual
periodicity
(p-value, R2)
0.739
0.704
0.002,
0.21
0.799
0.701
0.401
0.591
0.823
0.708
0.719
0.507
2
Autumn weather parameters
Explained variance by multivariate best fit models selected by stepwise regression
3
8
HFMD cases
Visiting hospital
5 cases each
week in each
province
5 cases each
month in
each county
May 2008 to
June 2009
Since June
2009
Severe cases
Mild cases
Samples
collected for
laboratory
test
Positive test
for enterovirus
Without
samples
collection
Samples collected
from each case
Laboratory test
for enterovirus
Negative test
for enterovirus
Negative test
for enterovirus
Defined as HFMD case with
any following complication:
- aseptic meningitis
- encephalitis
- encephalomyelitis
- AFP
- ANS dysregulation
pulmonary oedema
- pulmonary haemorrhage
cardiorespiratory failure
Positive test
for enterovirus
Probable cases
Lab-confirmed cases
Filling a form, collect epidemiological,
clinical and laboratory analysis data
Lab-confirmed cases
Reporting to HFMD
surveillance system
Appendix Figure 1. Flowchart of data and sample collection for HFMD cases, 2008-12, China
The enhanced surveillance system requires that each probable or confirmed case be reported on-line to the national CDC within 24 hours of
1
diagnosis. Data are updated as appropriate throughout the course of illness, and particularly at the conclusion of each episode. AFP=acute flaccid
paralysis. ANS=autonomic nervous system.
2
A
B
Appendix Figure 2. Map of the Chinese mainland provinces
31 Chinese mainland provinces were included in this analysis. The different colors
represent the different climatic or geographic regions. (A) Geographic regions in
Mainland China. Blue: mid-temperate region; green: warm-temperate region; black:
cold region; red: sub-tropic region; turquoise: tropic region. (B) Geographic regions in
Mainland China. Northeast: Liaoning, Jilin, Heilongjiang; North: Beijing, Tianjing,
Hebei, Shanxi, Inner Mongolia; Northwest: Shaanxi, Gansu, Qinghai, Ningxia,
3
Xinjiang; East: Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, Shandong;
Central: Henan, Hubei, Hunan; Southwest: Chongqing, Sichuan, Guizhou, Yunnan,
Tibet; South: Guangdong, Guangxi, Hainan.
4
5
Appendix Figure 3. Proportions of enterovirus serotypes among laboratory-confirmed HFMD cases by week of illness onset and by
geographic region, 2008-12, China
EV71=enterovirus 71. CV-A16=Coxsackievirus A16.
6
Figure 4. Age distribution and clinical severity of probable and
laboratory-confirmed HFMD cases, 2010-12, China
(A) Age distribution of probable and lab-confirmed cases. (B) Risk of fatality among
cases by age group and viral etiology. (C) Risk of severe illness among cases by age
group and viral etiology. (D) Risk of fatality among severe cases by age group and
viral etiology. EV71=enterovirus 71. CV-A16=Coxsackievirus A16. mth=months.
y=years.
7
Figure 5. Age distribution and clinical severity of probable and
laboratory-confirmed HFMD cases, 2012, China
(A) Age distribution of probable and lab-confirmed cases. (B) Risk of fatality among
cases by age group and viral etiology. (C) Risk of severe illness among cases by age
group and viral etiology. (D) Risk of fatality among severe cases by age group and
viral etiology. EV71=enterovirus 71. CV-A16=Coxsackievirus A16. mth=months.
y=years.
8
EV71 epidemics
CV-A16 epidemics
9
Other enterovirus epidemics
Appendix Figure 6. Latitudinal and longitudinal gradients in seasonality of EV71,
CV-A16 and other enterovirus
Figure shows that all enteroviruses exhibited latitudinal gradients in the amplitude of
annual and semiannual epidemics. Annual peak timing of CA-V16 activity occurred
earlier than for other enteroviruses (p=0·0148). Southern provinces experienced
earlier spring and fall epidemics of EV71 and other enterovirus than Northern
provinces (difference between the annual peak timing of epidemics in Northern and
that in Southern in EV71: p=0·031, in other enterovirus: p=0·21; difference between
the semiannual peak timing of epidemics in Northern and that in Southern in EV71:
p= 0·0008, in other enterovirus: p= 0·0008). The timing of fall EV71 epidemics was
also associated with longitude, with later occurrence in coastal provinces.
EV71=enterovirus 71. CV-A16=Coxsackievirus A16.
10
EV71 epidemics
CV-A16 epidemics
Other enterovirus epidemics
Appendix Figure 7. Periodicity of EV71, CV-A16 and other enterovirus epidemics in China
Importance of the semiannual periodicity, measured by the ratio of the amplitude of the semiannual cycle to the sum of the amplitudes of annual
and semiannual cycles. EV71=enterovirus 71. CV-A16=Coxsackievirus A16.
11
Appendix Figure 8. Importance of semiannual seasonality of EV71, CV-A16 and
other enterovirus
Importance of the semiannual periodicity, measured by the ratio of the amplitude of
the semiannual cycle to the sum of the amplitudes of annual and semiannual cycles.
EV71=enterovirus 71. CV-A16=Coxsackievirus A16.
12
A
B
C
Appendix Figure 9. EV71 epidemiological regions and predictors
(A) Identified epidemiological regions based on hierarchical clustering, using the
Euclidian distance between weekly standardized EV71 time series. Provinces are
color-coded by climatic region (see legends of appendix figure 1). (B) Map of the
three epidemiological regions identified in panel A. (C) Predictors of the two main
clusters identified in panel A, based on stepwise discriminant analysis.
13
A
B
C
D
14
Appendix Figure 10. CV-A16 epidemiological regions and predictors
(A) Identified epidemiological regions based on hierarchical clustering, using the Euclidian distance between weekly standardized CV-A16 time
series. Provinces are color-coded by climatic region (see legends of appendix figure 1). (B) Map of the three epidemiological regions identified
in panel A. (C) Climate predictors of the two main clusters identified in panel A, based on stepwise discriminant analysis. (D) Climate
predictors of the two sub-clusters identified in panel A, based on stepwise discriminant analysis.
15
A
B
C
Appendix Figure 11. Other enteroviruses epidemiological regions and
predictors
(A) Identified epidemiological regions based on hierarchical clustering, using the
Euclidian distance between weekly standardized other enterovirus time series.
Provinces are color-coded by climatic region (black: cold-temperate, blue:
mid-temperate, green: warm temperate, orange: subtropical, red: tropical). (B) Map
of the three epidemiological regions identified in panel A. (C) Climate predictors of
the two main clusters identified in panel A, based on stepwise discriminant analysis.
16